Multi-level multi-agent systems for traffic simulation: current trends and future challenges

Over the last few decades, multi-agent systems have been intensively deployed to simulate traffic. The transportation community has developed microscopic, mesoscopic, and macroscopic models to achieve a suitable trade-off between accuracy and reasonable computational burden. However, it has been recognized that none of these model types is superior to all others and capable of offering a single solution for largescale dynamic simulation to support traffic management needs. As the attention paid to large-scale traffic simulation continues to rise, the present paper offers a comprehensive review of multi-level multi-agent traffic simulation systems. The search for relevant publications in English was conducted, consulting four databases (Transport Research International Documentation (TRID), Scopus, ScienceDirect, and Google Scholar). Literature is compared in detail based on the research objectives and the design patterns to discuss the current trends and future challenges.